Vous êtes ici Ecole Doctorale Sciences Ecole Doctorale Thématique STAT-ACTU Agenda An introduction to Bayesian inference and flexible distribution

Outils

A venir...

Aucun événement

Offres d'emplois

Aucune offre d'emploi

Administration

An introduction to Bayesian inference and flexible distribution

Le Mercredi 4 Mars 2015 de 9h30 à 17h00
An introduction to Bayesian inference and
flexible distribution
by F.J.Rubio, Campus du Solbosch,
Batiment R42, Salle R.42.2.113
http://www.ulb.ac.be/campus/solbosch/plan-R42.html
4 March 2015, 9h30-12h30, 14h-17h
ULB

Contact person: Yves Dominicy (yves.dominicy@ulb.ac.be)
Registration is required (send an e-mail to yves.dominicy@ulb.ac.be).
There will also be a free lunch, but you need to register for it as well.
A certificate of attendance will be availablen if needed.

ABSTRACT :
In this short course we will study some principles of Bayesian inference and its use on some
classes of flexible distributions. We will start by pointing out some mathematical and philosophical
differences between the Bayesian paradigm and the Classical paradigm. After this, we will discuss
the crux of the Bayesian paradigm, namely, the choice of the prior. We will discuss both proper and
improper priors. Then, we will focus on a popular choice of the prior proposed by Harold Jeffreys
(The Jeffreys prior). The use of Markov Chain Monte Carlo (MCMC) simulation for approximating
the posterior distribution as well as other posterior quantities of interest will be briefly presented. The
second part of the course concerns the study of two families of flexible distributions: the family of
two–piece distributions and the family of skew–symmetric distributions. These kinds of distributions
are often used for modelling data that present departures from the assumption of normality in terms
of asymmetry and tail behaviour. We will finally present the corresponding Jeffreys priors for these
models and discuss the existence of the posterior distribution. If time permits, we will discuss some
general methods for generating new flexible distributions.

Dernière mise à jour par EDT STAT-ACTU Vendredi 16 Janvier 2015